The prevalence of passerines (mostly oscines, or songbirds) in international bird trade suggests that the possession or production of a song that is attractive or desirable to people may contribute to the likelihood of a species being traded. Testing this is difficult because we lack a general and readily available metric that quantifies attractiveness of bird song to humans. We propose and validate such a metric, based on the number of sound files lodged for a species on the Xeno-Canto website (www.xeno-canto.org). Our hypothesis is that species with more attractive songs are likely to be recorded more often, and so be represented more often in this online bird sound resource, all else being equal. Using a sample of North American and European passerines, we show that song repertoire size and geographic range size are consistently related to the number of recordings on Xeno-Canto. We use these results to derive a metric (the residuals of a model of the number of recordings in Xeno-Canto as a function of geographic range size) that may identify songs that are attractive to humans. Bird species whose songs are known to have inspired classical music, including several well known for their songs (e.g. common nightingale, European blackbird), have higher values of the metric than those that have not been referenced in classical music. The metric may help explain which bird species are present in trade, and so contribute to studies of invasion and conservation biology.